Real-Space Refinement

Real-space refinement in PrimeX makes use of Prime technology with OPLS force fields for minimization, loop building and side-chain placement, but with the addition of X-ray terms to the force field terms in the function to be optimized. The balance between the two can be set in the Calculation Settings Dialog Box. For information on Prime methodology, see the Prime User Manual — Contents.

Real-space refinement calculations are performed from the Real-Space Refinement Panel. To open the panel, click the task button for the relevant task in the Real-Space Refinement section of the PrimeX panel.

There are three choices of real-space refinement tasks:

• Refine loops
• Predict side chains
• Minimize

These tasks and other settings are described in the Real-Space Refinement Panel topic.

The Calculation Settings Dialog Box also has settings for the source of the map, which can be taken from a phased reflections file, and for the B-factors of the atoms to be fit in the real-space refinement.

Some notes on the tasks are given below.

Minimize

Real-space minimizations can sometimes result in nonplanar structures for groups that are nominally planar, such as peptide links, other amides, and unsaturated ring systems. The nonplanarity may appear to be large, but is usually not more than a few degrees. These small deviations from planarity may arise because of interactions with other parts of the structure or noise in the X-ray data. Other refinement programs enforce planarity for these groups quite strongly. Strictly enforcing planarity is in fact unphysical: these groups are not rigid, but the energetic cost for deviating from planarity is fairly high. PrimeX allows small deviations from planarity in an effort to provide the physically correct degree of restraint. Further refinement in PrimeX can sometimes alleviate the unfavorable interactions that resulted in the nonplanarity and return the groups to as near planar as is physically reasonable. If the cause of the deviation from planarity is due to noise in the X-ray data, decreasing the weight on the data relative to the restraining force fields will improve the results.

Predict side chains

Side-chain prediction is a necessary follow-up to mutation of the model to match the input sequence (which you must do before you can run a side-chain prediction). The mutation is performed automatically by the Mutate Model to Sequence task. To perform this task, first click Mutate Model to Sequence in the PrimeX panel, then in the Mutate Model to Sequence Dialog Box, click Mutate to Match Sequence. You can then proceed to side-chain prediction by clicking Place Side Chains, which opens the Real-Space Refinement panel with Predict side chains tools displayed.

You can also perform residue mutation manually, using the procedure described on PrimeX Basics: Modifying Structures, and adjust the side chain manually, using the Rotamers dialog box as described on PrimeX Basics: Structure Adjustments. However, this manual procedure only selects certain dihedrals, whereas the side-chain prediction varies the dihedrals in the side chain to find the best fit.

Side-chain prediction may be necessary after loop building, since the building process for the main chain does not comprehensively optimize the positions of the side chains. In addition, it can be useful at any other time during refinement to help improve side-chain placement in electron density.

The side-chain prediction task places all the selected side chains in a single run: there is no need to optimize them independently. Side chains are placed using a single omit map. Omission of more than 5% of the atoms in the map calculation can reduce the quality of the map and hence the accuracy of the side-chain placement. Thus, multiple side chains should be optimized at the same time, but selection of a large number of side chains carries some risk.